Emotion Corpus Study Using an Emotional Lexicon Dictionary in Design Scenarios -Focusing on the eVolo Architecture Competition Scenario-

서지정보

· 발행기관 : 한국공간디자인학회
· 수록지 정보 : 한국공간디자인학회 논문집 / 18권 / 6호 / 103 ~ 114페이지
· 저자명 : 황정석, 조택연

초록

(Research Background and Purpose) The linguistic understanding of Design is an important insight that can lead to the interpretation and creation of formal elements. Although early philosophy distinguishes between what can be said and what cannot be said, and what cannot be said should be kept silent, the interpretation of Design phenomena is possible through Design Scenarios, which are the spoken results of Design and formal elements. Moreover, both Design and language are embedded with emotions, and understanding the structure of emotions is important as a point of contact for all cognition. This study aims to improve the linguistic understanding of Design by structuring a Corpus of Design language derived from Sentiment Analysis, targeting the language of the eVolo architectural competition, which describes Design Scenarios for global conflict resolution and future spatial living. (Method) In order to recognize Design emotions in the language structure, we analyse the previous research on big data-based sentiment analysis and natural language processing. For the Sentiment Analysis experiment, the scenarios of 115 winning works from the five years of the eVolo competition are selected as the scope of the study. Analyse the emotions reflected in the language using the BERT and NRC emotion dictionaries. Classify Design emotions in a linguistic Corpus using the frequency of emotion words in the pre-processed data, and define a Design topic Corpus using topic modelling. By applying structured language research methods to Design research, we confirm the possibility of effective interpretation and creation through understanding the emotional structure implied in the Design of future architecture and space. (Results) This study used techniques from linguistic research to analyse and structure the complex emotional structure of the Design Scenarios. The resulting keywords can be organised into 9 emotion categories and 10 topic corpora, suggesting that in the future, Designs with similar emotional and topic structures can be structured based on language in Design and moulding production. (Conclusion) There's a lot to think about when it comes to why language is important in Design. As Ray Zekendorf said, nothing is communicated without language, we can clearly recognize the essence of meaning through language, and that is important. Moreover, in the era of realised generative AI, Designers also need considerable linguistic understanding for effective prompt input, and the attempt to recognize the emotional Corpus structure of Design languages through Sentiment Analysis presented in this study is effective. However, as this study focuses on the language used around the emotional structure rather than the specific formative language, the relationship between form and language needs to be more clearly supported by future research, such as the discovery of formative language dictionaries based on big data.

참고자료

· 없음